CBM3032 Mainboard Repair

[Wreck]

Hello,

My trustworthy 3032 developed several hiccups over the past months. Suddenly it wouldn't start, only with the help of the PET ROM/RAM board it would, but the Commodore text would be "wobbly" (with interference).
The system has the original white sockets installed. Applying a bit of pressure on UF10 the text would go from wobble to normal. After releasing pressure, the wobble would come back.
I concluded that the culprit was the sockets; I de-soldered the old ones and soldered new, good-quality sockets back in (UF10, UD9-UD2, UC7-UC4).

After putting everything back and powering on, nothing happened (except a nice and steady picture, but not the standard "garbage screen"). Even with the help of the PET ROM/RAM board, no change in behavior.

I measured VR3 and VR4; both output at 5.00V. VR5 is slightly less at 11.72V, don't know if this could be a problem.

PIN40 of the 6502 was found LO instead of HI; turned the machine OFF and measured R15,R16, C66-C68 and found all OK, so turned machine back ON and suddenly everything worked as it should. During loading a program from tape, the system froze; PIN40 was LO again.

I tested pins 1-8 on the 555 and got the following (with a logic probe)
A2 555
PIN1 LO
PIN2 LO
PIN3 HI
PIN4 HI
PIN5 NIL
PIN6 HI
PIN7 HI
PIN8 HI

A3 74LS04
PIN1 LO
PIN2 LO

A10 7417
PIN13 LO
PIN12 LO

I think further troubleshooting is needed; I believe the problem is somewhere in the timing logic(?).

Any ideas appreciated, thanks

Joe

 

[Wreck]

I desoldered the ceramic capacitor at C68 today, to have it tested on the bench, and thought "lets see what happens if I turn the machine on without this thing installed".
To my surprise the CBM works normally (it needs a push on the external reset button), PIN40 on the 6502 is HI and there is activity on PIN7.
I am currently running the PETTESTER in UD8 to see what it does.
I am a bit at loss regarding the C68 capacitor; it returns a 100nF value when testing with the multimeter. Am I missing something?

Thanks for any input

Joe

 

[daver2]

There is something not correct in your measurements (or reality):

A2 pin 3 (555 output) you measured as HI.

This is (or should be) directly connected to A3 pin 1. You measured this as LO. This can't be right...

A3 pin 2 (the output) should be the inverse of A3 pin 1 (the input). You measured LO again. This can't be right either...

It could be possible that you have misread A3 pin 1?

A2 pin 2 you measured as LO. This should be HI.

The 555 causes untold problems. Either the capacitors are leaking, the 555 is failing or there is flux on the solder joints that has gone partially conductive.

Pin 40 of the CPU being LOW (in a RESET condition) is not good and definitely indicates a problem with the reset circuit.

Dave

 

[Wreck]

There is something not correct in your measurements (or reality):

A2 pin 3 (555 output) you measured as HI.

This is (or should be) directly connected to A3 pin 1. You measured this as LO. This can't be right...

A3 pin 2 (the output) should be the inverse of A3 pin 1 (the input). You measured LO again. This can't be right either...

It could be possible that you have misread A3 pin 1?

A2 pin 2 you measured as LO. This should be HI.

The 555 causes untold problems. Either the capacitors are leaking, the 555 is failing or there is flux on the solder joints that has gone partially conductive.

Pin 40 of the CPU being LOW (in a RESET condition) is not good and definitely indicates a problem with the reset circuit.

Dave

So now with C68 removed, the CBM runs fine (needs a manual reset after power on). Starting from the 555 I now have the following;
A2
PIN1 LO
PIN2 HI
PIN3 LO
PIN4 HI
PIN5 HI
PIN6 HI
PIN7 LO
PIN8 HI

The inverter at A3 works at it should;
PIN1 LO
PIN2 HI

The buffer at A10 has both PINs 13 and 12 HI, therefore the 6502 PIN40 is also HI (and happy).

I had the PETTESTER running for some time now, it is doing its checks of the dram (pass 00002d at the moment), will have to turn off the machine for the night though.

What confuses me (as an absolute layperson) was the returned value of C68 with the multimeter; that seemed fine, but obviously it isnt. I will check tomorrow and putting back C68 and measuring the voltages on the 555; I am curious (as now PIN3 is in the mV range).
So far this has been a pretty steep learning curve with good success; it is very motivating to see things moving in the right direction.

Question; can I test those ceramic capacitors with one of these fancy ESR meters?

All feedback greatly appreciated

Joe

 

[daver2]

It is just easier to replace the ceramic capacitor... It is cheaper and less stressful (unless you already have the necessary test equipment)...

It could still be the 555 or some trapped 'gunge' (perhaps on the PCB under the 555) that is aging and starting to go conductive.

These machines were never designed to run this long, and PCB cleanliness was not high on the list of priorities after it had been manufactured.

We have seen a couple of PETs where replacing every component in the reset circuit (555, resistors, capacitors) did not fix the problem. Only scrubbing the PCB (top and bottom around the area occupied by the 555) with isopropyl alcohol appeared to fix the problem...

Dave

 

[Wreck]

It is just easier to replace the ceramic capacitor... It is cheaper and less stressful (unless you already have the necessary test equipment)...

It could still be the 555 or some trapped 'gunge' (perhaps on the PCB under the 555) that is aging and starting to go conductive.

These machines were never designed to run this long, and PCB cleanliness was not high on the list of priorities after it had been manufactured.

We have seen a couple of PETs where replacing every component in the reset circuit (555, resistors, capacitors) did not fix the problem. Only scrubbing the PCB (top and bottom around the area occupied by the 555) with isopropyl alcohol appeared to fix the problem...

Dave

Hi Dave,

I will clean the board again to make sure no gunk remained from previous repairs; somebody soldered new 2114 sockets in, and even I am quite a layman, can say the job doesnt look great (no impact on function).

I will try and replace the capacitor, have got tons of ceramic capacitors around but they all seem to be way smaller than the original one so I need to probably source from my local shop.

Just out of curiosity; how long does the PETTESTER routine normally run before it is done? I would like to give the machine a full test run before putting it back into the case. I have another board waiting for some repairs...

Joe

 

[daver2]

The problem with the reset circuitry is 'detritus' around the 555 of course. However, detritus elsewhere can also spell trouble for later on.

I normally clean everything up and then look at my work with a 'critical hat' on...

The PETTESTER just sits in the DRAM test for eternity...

I would normally give it a count of 20 (hex) or so before giving the DRAM a pass...

Dave

 

[Wreck]

After fixing the reset circuit, havin a nice and stable platform, I thought I will hook up a cassette tape and try to load a test program from tape to see how things are with loading stuff from external sources.

Soon after the tape was searching for the program, the picture started to wobble again and random characters appeared.

After turning everything off and back on, there was no picture.
I discovered there is no CLK signal on PIN37 of the 6502, and the relevant pins on J7 had no HORZ / VERT frequency.

Any ideas appreciated what could have caused such a dramatic effect ...

Thanks

Joe

 

[Wreck]

OK So I traced the problem all the way back to the 16MHZ crystal Y1. for some reason or the other it is not producing any signal. I suspect C44 being faulty, but need to confirm once I have a bit more time.

 

[daver2]

Crystals go faulty on a regular basis (if it is not the capacitor).

This is usually what happens with vintage equipment that you are trying to resurrect!

Dave

 

[Wreck]

It is definitely helpful to have a large cupboard with plenty of spares, but probably more important to have an understanding wife. Going to get some 22pF Capacitors tomorrow...unfortunately they don't look original.

 

[Hugo Holden]

Crystals do deteriorate over time and they age and it is affectionately called, losing activity. In many cases though it does not prevent oscillations. However, many logic circuits that use crystals deploy a sort of fudge to convert a digital device like an inverting gate, into an analog amplifier, by connecting a resistor from its input to its output. Then two of these are strung into a loop to have non-inverting gain with the crystal in the loop, and the positive feedback is enough to sustain oscillations, even with the losses from the crystal. Using CMOS gates with high value feedback resistors such as 1 Meg Ohm are probably better than TTL gates requiring low value feedback resistors and they create a much lower input impedance "amplifier". Though, despite this they still work well with 74 series or 74LS series TTL gates.

When the situation goes borderline though, with aging of the crystal, you can get a scenario where it might start on one occasion and not on another. Temperature and vibrations can determine if it starts or not, also testing it with meters and scopes can trigger to start, also the rate that the power supply voltage rises at power up can determine if it starts too. So, usually before the crystal dies completely, there is a phase of intermittent starting behavior. Though, once started, they normally keep running even if marginal.

 

[Wreck]

Thanks, Hugo, for your very valuable input. I didn't have time yet to check further, however my 22pf capacitors arrived. At the next suitable occasion I will continue with the troubleshooting. What confuses me mildly are the two resistors R7/R8 in the oscillator loop which are, according the schematic 470 ohm rated, but on the parts list it states 270 ohms (unless I am missing something). Why would there be different values?

Thanks

Joe

 

[Hugo Holden]

Thanks, Hugo, for your very valuable input. I didn't have time yet to check further, however my 22pf capacitors arrived. At the next suitable occasion I will continue with the troubleshooting. What confuses me mildly are the two resistors R7/R8 in the oscillator loop which are, according the schematic 470 ohm rated, but on the parts list it states 270 ohms (unless I am missing something). Why would there be different values?

Thanks

Joe

Possibly the difference could be that the 270R resistors were for a plain 74 series inverter gate like a 7404, but a gate like a 74LS04, would use 470R because that gate's input currents are lower. Even though the inverter gates are biased as an analog amp, they still require enough input current to get their outputs to switch between close to zero and about +4v.

For example a typical Xtal osc circuit, using the 7404 IC would be expected to have about a 330R resistor from each gate input to its output.

Though a circuit using a 74LS04 it might be as high at 470R to 1K.

In most cases any feedback resistor in the range of 270R to 470R would work for most TTL IC's, but you would get a higher gain and more chance of it oscillating the higher the value of the resistor. But for LS TTL, not much above 2k would help, even for LS TTL, because higher values cannot pull the input pin reliably to logic low, because of the current sourced by the input pin.

For typical LS parts the pin sources about 0.36 mA. So to pull that low with a resistor, a 2k would drag it down to 0.72V, starting to get a little risky for a definite logic low and it would fail for a plain TTL with about 1.6mA current and only pull it down to 3.2V. But a 330R for example will pull a 74 series input pin down to 0.53v which is a solid enough logic low.

In CMOS systems it can be as high as 1 Meg Ohms, but that is too high to bias the TTL logic gate into a class A style analog amplifier condition.

A 470R is in a sort of no man's land for plain TTL. For 74LS TTL it is fine. In the case of plain TTL it can only lower the input pin voltage to 0.75V, which is a logic low, but with not a large margin. This is probably why most designers uses a 330R for a plain 74TTL osc circuit, but a 270R would work, but it could require a slightly higher activity crystal.

 

[Wreck]

Possibly the difference could be that the 270R resistors were for a plain 74 series inverter gate like a 7404, but a gate like a 74LS04, would use 470R because that gate's input currents are lower. Even though the inverter gates are biased as an analog amp, they still require enough input current to get their outputs to switch between close to zero and about +4v.

For example a typical Xtal osc circuit, using the 7404 IC would be expected to have about a 330R resistor from each gate input to its output.

Though a circuit using a 74LS04 it might be as high at 470R to 1K.

In most cases any feedback resistor in the range of 270R to 470R would work for most TTL IC's, but you would get a higher gain and more chance of it oscillating the higher the value of the resistor. But for LS TTL, not much above 2k would help, even for LS TTL, because higher values cannot pull the input pin reliably to logic low, because of the current sourced by the input pin.

For typical LS parts the pin sources about 0.36 mA. So to pull that low with a resistor, a 2k would drag it down to 0.72V, starting to get a little risky for a definite logic low and it would fail for a plain TTL with about 1.6mA current and only pull it down to 3.2V. But a 330R for example will pull a 74 series input pin down to 0.53v which is a solid enough logic low.

In CMOS systems it can be as high as 1 Meg Ohms, but that is too high to bias the TTL logic gate into a class A style analog amplifier condition.

A 470R is in a sort of no man's land for plain TTL. For 74LS TTL it is fine. In the case of plain TTL it can only lower the input pin voltage to 0.75V, which is a logic low, but with not a large margin. This is probably why most designers uses a 330R for a plain 74TTL osc circuit, but a 270R would work, but it could require a slightly higher activity crystal.

Dear Hugo,

Thank you for explaining, this actually makes sense to me!
I will have a few hours today to work on the board, and will check again the crystal circuit and the 7404 voltages. Maybe I will poach the crystal from another board and see if that works or if there is an issue elsewhere.

joe

 

[Wreck]

OK so the crystal works now; I have a 16MHZ clock going into I1, further into G1, G5 and into H3.
But there it stops. I should have an output on G5 on PIN7, but there is nothing.

Further trouble shooting in progress, any inputs highly appreciated

joe

 

[daver2]

I am a bit confused by your logical progression around the circuit...

The 16 MHz clock should go from G1/6 to H3/8. Is that present?

The 16 MHz clock should also go from G1/6 to G5/14. Is that present?

Next, I would check G5/11. This pin should be permanently HIGH.

Then check the outputs of G5:

G5/3 for an 8 MHz clock.
G5/2 for a 4 MHz clock.
G5/6 for a 2 MHz clock.
G5/7 for a 1 MHz clock.

Dave

 

[Wreck]

Hi Daver,

I will check tonight! Thanks!

Joe

 

[Wreck]

I am a bit confused by your logical progression around the circuit...

The 16 MHz clock should go from G1/6 to H3/8. Is that present?

The 16 MHz clock should also go from G1/6 to G5/14. Is that present?

Next, I would check G5/11. This pin should be permanently HIGH.

Then check the outputs of G5:

G5/3 for an 8 MHz clock.
G5/2 for a 4 MHz clock.
G5/6 for a 2 MHz clock.
G5/7 for a 1 MHz clock.

Dave

So I checked; the 16MHz are present all the way from G1/6 to H3/8 and to G5/14.
G5/11 is LO,
G5/3-7 no clock (well 50Hz it says).

Now I think next would be investigating why G5/11 is LO...

Joe

 

[daver2]

Ah, that signal is fundamental!

There should be a pull-up resistor (R12 = 1 k) that should pull this signal up to +5V.

Dave